PC vs PC defensive parses (long!)

[TerjynPovar]

Yes, indeed, I did miss that Coprolith stated in the first post that the soft cap varied by mob, my bad.  But I'm still glad I asked the question cause the followup post was very interesting.

I wish I had the patience you have Coprolith.

[Helben]

Soft cap has to be set independently on each mob. A fine example would be Innoruk and Rallos Zek in Time

Both are of same level, and both has about the same max hit. The problem is that RZ can and will hit your tank for max 1 or more hits every round while it can go several round without Inno hitting your tank for max. When RZ takes some pleasure hitting me, he usually max on all 4 hits it takes to kill me. Inno seldom hits for max. When he has taken the pleasure to hit me, I usually have lasted 7-8 hits before he kills me.

Btw, from the formulas that SoE has published about skill ups, I hardly think that we will find a universal formula for hit ratio and damage. For those not reading the tradeskill boards, the skill up formula is actually two tests with two different formulas, and if you pass both tests when using a skill, you get a skill up. The basics of the formulas are that you skill up more on success than on fail, you have higher chance of skilling up on fails if you have a high stat, it is harder to skill up with higher skills, and that one of the formulas has an unknown X that changes with the skill in question. It would not surprise me at all if they used multiple formulas in combat too, and that the formulas include some X that changes with the circumstances.

Cleben Lege, archon of Istid, AB

[Romidar]

Thanks for the very detailed parsing and analysis. What you post with respect to LR5 and ID5 is pretty much dead on to what we've discussed on Paladins of Norrath, but the data I collected came from actual fighting instead of controlled parsing. (That is, we found that ID5 was not nearly as good as LR5 at reducing damage.)

Interesting (and not surprising) to see the very tiny effect of agility. What makes me go "hmmm" is that the magnititude of the effect you observed there is of the same sort of magnitude that we observed for the effect of strength on damage output years ago - so tiny that you need a huge and change and a big sample to see it. In my parses of fighting with and without Ferine Avatar, it's easy to observe the offensive increase due to that buff (damage went way up) but almost impossible to see any change defensively.

We've known for a while that defensive skills are checked before hit/miss (e.g., riposte damage does not go down when your avoidance goes up), but it's always nice to see it measured again.

Thanks again for the data - I'll post a link to this on PoN if one hasn't been posted already. :)

[Coprolith]

I know that for people who've been around long enough and have been following the parsings over the years there won't be a lot of news in the parses; they can consider it a summary  :) . This is actually the point; the results are very similar to what has been parsed against real mobs, even tho the mitigation and avoidance numbers themselves have virtually no bearing on everyday EQ life. Classes for which tanking is their business (knights and warriors) would probably like to see numbers that are more applicable to real situations, but if your interest is in the game mechanics then parsing against PCs is, in my opinion, superior.

I started this because i got frustated about the tediousness of getting good enough statistics in defensive parses against NPCs. About a year ago i tried to parse out CS and CA during the aaxp grind and simply couldnt get proper results even after thousands of lines (mitigation going down 4% going from CS1 to CS2 with 4k samplesizes). The statistical variation during normal xp sessions must get magnified by factors from the combined effects of stuns, lag fading buffs etc. I then tried parsing CS2->CS3 using a single mob, without killing it, but my brain started dribbling out of my ears after just an hour. Its no surprise that high quality defensive parses are few and far between. Those people on the SW board who did the parses i mentioned must have incredible patience.

Anyway, I hope others will pick up on this method instead of just dimissing it as 'not realistic' so that we'll see more defensive parses in the future.

/hugs

[Moghedred]

Does this mean that agility affects mitigation as well as avoidance? No, its too early to tell from this data alone.

Hi Coprolith. Very interesting post there. I would like to add that Magelo calculates Avoidance and Mitigation AC separately even though it shows just one all-inclusive number.

According to their formula, AGI does indeed seem  to affect both Avoidance and Mitigation AC. I also noted that they seem to calculate that your base AGI (i.e. naked) has a larger impact on AC than worn AGI. But in the end I think it confirms what you say. See below!

Here's what Magelo spits out for me:

AC Formula from Magelo:                     
   Agi   T   Items   Avoidance   Mitigation    Total    +AC
Base   90   42   0   490   84   677   677
Base + mod1   90   42   432   490   660   1357   680
Base + mod2   178   58   0   506   84   696   19
--------------------------------------------------------------------------------

Total AC from formula: 1376
Correction value for AC: 10

--------------------------------------------------------------------------------

TOTAL AC 1386


To see this just type "&debug=true" after your profile URL.

edit: just can't get that table to format right... sorry did some mroe testing on Magelo.

So: from left to right you get AGI (total) then T Value then Items, Avoidance then Mitigation then Total then AC added.

Top to bottom you have base (I assume this is naked + my base AGI is indeed 90 and my fully+geared AGI is 178). Mod 1 and then Mod2. The worn AGI is Mod 2 and as you can see it adds nothing to the mitigation AC.
However, the Base AGI does appear to add 84 to my Mitigation AC. That's how I read it anyway. However, testing this by buying the +Agility AAs in Magelo does not change the base mitigation AC and only very slightly raises Avoidance (even at base).
Also note that Base AC is 490 Avoidance and 84 Mitigation with 90 AGI. With almost double that in fully-geared AGI I seem to only get an additional 16 in Avoidance AC and nothing in Mitigation (after buying AGI AAs it's only 15 added).  

So, I'm guessing that the base mitigation AC is derived from your defense skill or such.

Tested in Magelo: set Defense to 0 in Magelo and you get 0 base Mitigation AC. And also most of your base Avoidance goes. Your left with just avoidance of 48 at 120 base AGI. That suggests you need 2.5 AGI per base AC (Avoidance only) and 5.8 AGI per AC gained for worn AGI.

This also solves the T Value riddle. It's just the Avoidance AC gained from AGI.

Hope this helps!

[Coprolith]

Its not so much a difference between base agility and agi gained from items, its just that the effect of agility isnt constant. You get a bonus to your AC (which is pure avoidance AC) for having agi over 75, which also depends on your level. After that avoidance AC increases steadily by about 1 pt per 4.5 AGI up to 200AGI, and 1 pt per 20agi above 200.
Since most people will have their base AGI at or above 75 this makes it look effectively like base AGI gives more avoidance per point. But its really just your actual AGI that determines your bonus to AC (if your agi drops below 75 because of encumbrance or getting low on health you loose everything, not just the part you gained from worn AGI).

Basically, Magelo takes your avoidance AC from this graph:




/hugs

[Moghedred]

Nice! thanks a lot

/wave

[Coprolith]

Some interesting info from Kavhok (one of the EQ devs) on the Steel Warrior board (in this thread)

The cap on AC in the Velious era wasn't a soft cap; it was a hard cap that had been there from day 1. After a certain point, which differed for each class, the benefit of more AC didn't just diminish - it dropped to nothing.

The change I referred to, just before PoP, changed that from a hard cap to a soft cap. You get a percentage of the amount over that soft cap. Shields increase both your total and your soft cap, making them more effective than any other item with equal AC. Your mitigation AAs, level, and class also affect the cap and the percentage return for AC over it.

Separate from this, there are diminishing returns if your AC is much greater than the NPC's attack. This is due to the nature of the formulas that produce the probability distributions

3) if mitigation AAs are figured into the AC softcap, do you need a certain level of AC before you get full benefit from the mitigation AA's?

You folks do a great job of parsing and figuring out the mechanics of the game. I don't want to leave you with no mysteries at all =) I will say this:

3) Yes, the mitigation AAs will not do anything for you if your AC isn't high enough. However, the median level 65 warrior is over the cap even with both Luclin and PoP mitigation AAs maxed, so you can safely assume you're getting the full benefit out of them.

The interesting part about this is that the mitigation AAs actually increase your AC softcap, and that you don't get full benefits from them until you've reached that cap. What is true for the median warrior prolly wont be true for the median BL. I did my parses PvP against a L65 druid, for which ~1000AC was enough to reach the softcap w/o ID5. Purchasing ID5 might have put me over the cap (since i was just on the edge of the softcap before that), which would explain why i didnt get the same % increase as from LR5.
Something to remember: CS and ID get better as your AC gets higher. If your AC isnt high enough to reach the soft cap against the mobs you usually fight against, then getting CA/LR will be a lot better then CS/ID.

/hugs

[Toln]

Coprolith, is it your opinion that the soft cap returns we see on AC are entirely specific to each NPC and its ATK, or are there other factors too such as the player's Class?

I ask, because I stumbled across this detailed parse posted on the Monk forums which suggested that different classes all have different "soft AC caps", which are independent of the NPCs they are fighting.

http://p201.ezboard.com/fmonklybusiness43508frm2.showMessage?topicID=1259.topic

The parsed commentary seemed to indicate that two classes with identical avoidance/mitigation AC, such as Beastlord vs Warrior, would mitigate damage much differently.

Prior to reading that I was under the impression that the big difference between plate tank mitigation and leather/silk caster mitigation for example, was simply due to the itemization and difference in AC bonuses for the classes... resulting in those non-tank classes having a much harder time getting their AC high enough to mitigate effectively, but in theory if a person could get it up there high enough, they could tank just as good as a plate class with the same AC rating.
Is this incorrect? And if so, do you have any theory on what other game mechanics are affecting it?

[Coprolith]

It is entirely possible that the soft cap is class dependent as well as mob dependent. Or it could be that different classes are on different mitigation tables, that is, the AC soft cap is the same for all of them, but the level of mitigation above the soft cap is different. The parses on the MB board do not distinguish between these 2. If we believe the quote of the Dev in my prev post however then both are true: "Your mitigation AAs, level, and class also affect the cap and the percentage return for AC over it."
We also know that SOE can program specific bonuses to mitigation for certain classes as in the case of warriors. And finally, we now also know that the AC from a shield is also added to the soft cap.

That's a lot of variables in total. You'd need to parse extensively over a whole range of AC values for each class and for each variable to get a complete picture. With defensive parses being the chore they are, that's not going to happen anytime soon.

/hugs